Data were from the MCS (42–44). Enrollment methods have been previously described (45, 46). Briefly, service members were randomly selected from Defense Manpower Data Center (DMDC) rosters and invited to enroll in five panels in 2001–2003, 2004–2006, 2007–2008, 2011–2013 and 2020–2021 with over 260,000 participants enrolled to date. Follow-up surveys with questions covering physical and mental health, health behaviors, and life experiences were sent to enrolled participants every 3–5 years (43). Throughout this manuscript, we refer to the survey cycles by the year they closed (e.g., 2013 for the 2011–2013 cycle).

Figure 1 depicts the flow chart of inclusion criteria for the present research. MCS participants from Panels 1–4 who served on active duty and completed the 2013 survey, either as baseline or follow-up, were eligible for inclusion (n = 97,033). Participants were excluded if responses to the blast screening questions assessed on the 2013 survey (n = 6,379) or covariates of interest (n = 19) were missing. Because the primary focus of the present analyses was on medical diagnoses recorded in the VHA system, the present analyses were restricted to veterans who utilized VHA care (i.e., were seen for at least one healthcare visit for two or more calendar years; n = 36,641). Supplemental analyses were also conducted for the broader population of veterans enrolled in the VHA regardless of the number of visits over time (n = 51,541). Individual analyses for each condition of interest were further limited by excluding participants diagnosed with the condition in the VHA prior to the 2013 survey, which resulted in varying analytic sample sizes for each outcome of interest.

Descriptive statistics were calculated for military and demographic characteristics overall and stratified by occupational risk of LLB and self-reported HLB exposure. The geometric mean and standard deviation were calculated for the number of days deployed because this continuous variable displayed a log-normal distribution (51). Collinearity among covariates was examined with a variance inflation factor threshold of 4. For each condition, we calculated the numbers of cases diagnosed: (a) overall [i.e., in either the MDR or VHA medical systems], (b) in the MDR before the 2013 survey, (c) in the VHA before the 2013 survey, and (d) in the VHA after the 2013 survey. Although counts and frequencies are presented for all four criteria, this report will focus on cases identified in the VHA during the study period. Adjusted models were calculated excluding participants diagnosed in the VHA before the 2013 survey. Prevalence ratios for HLB exposure and LLB occupation for each condition were estimated using Poisson regression in the GENMOD procedure with the robust standard errors (52). Models adjusted for military (service branch, total days deployed, pay grade, deployment/combat experience) and demographic factors (age, sex, race, ethnicity, education, marital status), enrollment panel, diagnosis of the condition in the MDR before the 2013 survey, and the interaction between HLB and LLB. When models failed to converge due to a small number of participants with a prior diagnosis of the condition in the MDR before the 2013 survey (i.e., for memory loss and delirium/dementia), cases with prior history were excluded from adjusted analyses. False discovery rate (FDR) adjustment for multiple comparisons was used to identify significant direct associations and interaction effects (53). FDR corrections were chosen instead of traditional family-wise error rate corrections as they confer increased statistical power, which was deemed appropriate considering the smaller sample size than similar published investigations (15). When the two-way interaction variable between HLB and LLB was statistically significant (FDR corrected p < 0.05), adjusted associations were calculated with a common reference category of blast-naïve (i.e., those with no HLB and low risk of LLB exposure). Although we did not set a priori criteria for number of cases to conduct analyses, some models failed to converge, therefore adjusted prevalence ratios were not calculated. All analyses were conducted in SAS 9.4.

Table 1 reports sample characteristics for the full sample and by occupational risk for LLB and self-reported HLB exposure. Among the analytic sample of veterans who utilized VHA care for two or more years during the study period (N = 36,641), most participants were enrolled in the 2013 survey, male, identified as White/non-Hispanic race and ethnicity, were born after 1979, attended some college but had not obtained a degree, were married, were enlisted, and served in the Army (Table 1). The average total time deployed was 302 days (geometric SD = 2.49). Approximately 17.4% of the sample worked in an occupation at high risk for LLB exposure; the most common high-risk occupations included general infantry (37.9%), combat operations control (12.6%), and combat engineering (8.1%). In addition, 10.6% of the sample reported HLB exposure on the 2013 MCS survey. All military and demographic characteristics examined were significantly different by occupational risk of LLB exposure and self-reported HLB exposure.

Supplementary Tables 2–4 display results among veterans enrolled for two or more years in the VHA during the study period (N = 51,541), regardless of utilization. Previous reported findings included only VHA utilizers (i.e., those enrolled in the VHA with at least one medical encounter per year for 2 years). Case counts mirrored those observed in the main analyses although rates of conditions were generally lower, consistent with the larger sample size that includes VHA nonusers. Adjusted model results are consistent with a few exceptions (Supplementary Tables 3, 4). Among TBI-related conditions, HLB was additionally associated with sleep disruption movement disorders and communication disorders, whereas LLB was additionally associated with migraine headaches and non-headache pain. The interaction between HLB and LLB was significant for four additional conditions: skin sensation disturbances, non-headache pain, syncope and collapse, and cognitive problems. Like the primary analyses, these interaction effects were antagonistic. Among mental health diagnoses, HLB was associated with acute stress disorder and LLB was additionally associated with depressive disorders.

Although previous research has documented associations among HLB and LLB on long-term health outcomes, there is a paucity of work evaluating health care utilization within the VHA due to BOP exposure. These results underscore the strong association between blast exposure and adverse health outcomes, particularly the high magnitude of risk observed for TBI, memory loss, and post-concussive syndrome. Veterans exposed to blasts during their military service may continue to face significant repercussions linked to BOP exposure. The conditions examined herein (e.g., TBI, mental health conditions) represent significant burden on health care systems both for the direct clinical management of the condition and the secondary effects on physical health (57, 58). These results highlight that the long-term effects of BOP exposure on physical health and mental health represents a concern not only for the DoD, but also for entities such as VHA that provide care for veterans.

Regarding military health policy and preventive care, our results highlight the need for systematic tracking and planned analysis of both HLB and LLB exposures over the course of a service member’s career. Incorporating blast exposure histories—using tools such as MOS risk categories or standardized self-report instruments—into routine medical records could help the DoD and VHA better anticipate long-term healthcare needs, enable early screening for at-risk individuals, and guide the development of targeted rehabilitation and clinical monitoring protocols. Further, incorporating blast exposure into disability evaluations for VHA should be considered in future. From a policy standpoint, this study supports the implementation of occupational blast exposure surveillance systems and the development of evidence-based exposure thresholds that inform safe training practices, consistent with recent DoD policy (59).

Understanding the full implications for treatment and management of these conditions requires a better understanding of the potential mechanisms (e.g., neurological, psychological) by which BOP exposure may increase likelihood of diagnosis for the observed conditions. Possible effects of BOP exposure on brain structure and function have been previously documented (12, 13, 19, 26, 39, 60, 61). Previous research has noted the association between TBI and psychological health, though it remains unclear whether this is due to structural or functional damage to the brain or psychological mechanisms due to a traumatic event (62).

Of particular concern for preventive efforts are the outcomes of LLB including PTSD, adjustment disorder, hearing conditions, and cognitive problems. These associations were independent of HLB, suggesting the possibility that service members chronically exposed to LLB during training activities may have an increased likelihood of these disorders in the future. This is further supported by findings that chronic LLB exposure, rather than acute LLB exposure, is typically associated with increased symptom reporting (33, 35, 38, 39).

The presented study results need to be considered within the context of limitations. First, HLB was self-reported. Service members reported whether they had been exposed to a blast/explosion (i.e., HLB), but were not asked to provide additional information regarding characteristics of that HLB exposure (e.g., distance, intensity, frequency). Additionally, it was not possible to directly link medical record data to the individually reported HLB events, which limits the ability to verify the severity of specific exposures.

Low-level blast risk was estimated using MOS as a proxy for exposure risk. This method is consistent with prior work (2, 15, 47, 63) and allows for large-scale estimation, but it does not capture individual variation in cumulative exposure, time in service, or specific training environments. However, prior work supports the validity of MOS as a proxy, showing that MOS-based classifications correspond to actual exposure experiences (47). Although imperfect, this method provides a reasonable and validated strategy for modeling LLB exposure in large cohort studies, including the MCS (1, 15). In addition to its utility in epidemiological research, MOS may also hold practical value for identifying individuals at elevated risk for blast-related brain health outcomes. Recently directed tracking of BOP (59) incorporates MOS-based risk stratification into military and post-service healthcare systems which is intended to support targeted mitigation strategies during training (e.g., modified exposure limits, protective equipment), and early identification of individuals who may benefit from clinical monitoring or intervention. This approach is expected to enhance prevention and treatment efforts by allowing the MHS and VHA to proactively address the needs of service members and Veterans most likely to exhibit blast-related health concerns. The continued improvement of objective tools to measure exposure, such as body-mounted sensors, biometric wearables, and biospecimen data, may improve the precision of BOP exposure estimates. Although practical limitations currently restrict the broad application of these tools, they have strong potential to complement self-report and administrative data in future research.

Second, our primary analytic sample focused on veterans who had been identified as VHA users (i.e., enrolled in the VHA with at least one medical encounter per year for at least 2 years), meaning all participants had to meet eligibility criteria that typically require documented health concerns related to military service. Although this sample may be representative of VHA users broadly, generalizability to veterans who do not regularly use or are not eligible for VHA services may be limited (64–66). In addition, there is potential for collider bias, as exposures to blast and experience of conditions may result in a greater likelihood of veterans using VHA care. Veterans from lower-risk occupational backgrounds would therefore likely only be represented in this sample if they exhibited sufficient service-related symptoms to access care, which may bias the sample toward more symptomatic individuals in this group. Regardless, our results provide a valuable source of information regarding cost burden to the government for continued healthcare services following separation from service.

Finally, the analyses from the present research were limited by features of the study design. For example, we were unable to evaluate whether TBI may have moderated the associations between HLB and LLB and other subsequent health outcomes. Detailed TBI data were not available in the 2013 survey cycle, therefore we were unable to determine if TBI diagnoses occurred concurrent to reported HLB. Additionally, not all VHA medical records included diagnosis order, therefore we were unable to apply exact AFHSB case definitions that require the ICD code to be the first or second diagnostic code.

Despite these limitations, the present research has notable strengths including its use of a large, representative sample and prospective design. We were also able to account for several demographic and military factors, including deployment with and without combat experience. These strengths, combined with the ability to link self-report data with VHA medical records, enabled one of the first prospective examinations of long-term health outcomes associated with BOP exposure to date.